Muse: Towards Reproducible Long-Form Song Generation with Fine-Grained Style Control

• URL: http://arxiv.org/abs/2601.03973v3
• Date: Mon, 12 Jan 2026 11:39:42 GMT
• Title: Muse: Towards Reproducible Long-Form Song Generation with Fine-Grained Style Control
• Authors: Changhao Jiang, Jiahao Chen, Zhenghao Xiang, Zhixiong Yang, Hanchen Wang, Jiabao Zhuang, Xinmeng Che, Jiajun Sun, Hui Li, Yifei Cao, Shihan Dou, Ming Zhang, Junjie Ye, Tao Ji, Tao Gui, Qi Zhang, Xuanjing Huang,
• Abstract summary: We release a fully open-source system for long-form song generation with fine-grained style conditioning.<n>The dataset consists of 116k fully licensed synthetic songs with automatically generated lyrics and style descriptions.<n>We train Muse via single-stage supervised finetuning of a Qwen-based language model extended with discrete audio tokens.
• Score: 66.46754271097555
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Recent commercial systems such as Suno demonstrate strong capabilities in long-form song generation, while academic research remains largely non-reproducible due to the lack of publicly available training data, hindering fair comparison and progress. To this end, we release a fully open-source system for long-form song generation with fine-grained style conditioning, including a licensed synthetic dataset, training and evaluation pipelines, and Muse, an easy-to-deploy song generation model. The dataset consists of 116k fully licensed synthetic songs with automatically generated lyrics and style descriptions paired with audio synthesized by SunoV5. We train Muse via single-stage supervised finetuning of a Qwen-based language model extended with discrete audio tokens using MuCodec, without task-specific losses, auxiliary objectives, or additional architectural components. Our evaluations find that although Muse is trained with a modest data scale and model size, it achieves competitive performance on phoneme error rate, text--music style similarity, and audio aesthetic quality, while enabling controllable segment-level generation across different musical structures. All data, model weights, and training and evaluation pipelines will be publicly released, paving the way for continued progress in controllable long-form song generation research. The project repository is available at https://github.com/yuhui1038/Muse.

Related papers

• Music Boomerang: Reusing Diffusion Models for Data Augmentation and Audio Manipulation [49.062766449989525]
  Generative models of music audio are typically used to generate output based solely on a text prompt or melody.<n>Boomerang sampling, recently proposed for the image domain, allows generating output close to an existing example, using any pretrained diffusion model.
  arXiv  Detail & Related papers  (2025-07-07T10:46:07Z)
• SongGen: A Single Stage Auto-regressive Transformer for Text-to-Song Generation [75.86473375730392]
  SongGen is a fully open-source, single-stage auto-regressive transformer for controllable song generation.<n>It supports two output modes: mixed mode, which generates a mixture of vocals and accompaniment directly, and dual-track mode, which synthesizes them separately.<n>To foster community engagement and future research, we will release our model weights, training code, annotated data, and preprocessing pipeline.
  arXiv  Detail & Related papers  (2025-02-18T18:52:21Z)
• MusiConGen: Rhythm and Chord Control for Transformer-Based Text-to-Music Generation [19.878013881045817]
  MusiConGen is a temporally-conditioned Transformer-based text-to-music model. It integrates automatically-extracted rhythm and chords as the condition signal. We show that MusiConGen can generate realistic backing track music that aligns well with the specified conditions.
  arXiv  Detail & Related papers  (2024-07-21T05:27:53Z)
• MuPT: A Generative Symbolic Music Pretrained Transformer [56.09299510129221]
  We explore the application of Large Language Models (LLMs) to the pre-training of music. To address the challenges associated with misaligned measures from different tracks during generation, we propose a Synchronized Multi-Track ABC Notation (SMT-ABC Notation) Our contributions include a series of models capable of handling up to 8192 tokens, covering 90% of the symbolic music data in our training set.
  arXiv  Detail & Related papers  (2024-04-09T15:35:52Z)
• JEN-1 Composer: A Unified Framework for High-Fidelity Multi-Track Music Generation [18.979064278674276]
  JEN-1 Composer is designed to efficiently model marginal, conditional, and joint distributions over multi-track music.<n>We introduce a progressive curriculum training strategy, which gradually escalates the difficulty of training tasks.<n>Our approach demonstrates state-of-the-art performance in controllable and high-fidelity multi-track music synthesis.
  arXiv  Detail & Related papers  (2023-10-29T22:51:49Z)
• Simple and Controllable Music Generation [94.61958781346176]
  MusicGen is a single Language Model (LM) that operates over several streams of compressed discrete music representation, i.e., tokens. Unlike prior work, MusicGen is comprised of a single-stage transformer LM together with efficient token interleaving patterns.
  arXiv  Detail & Related papers  (2023-06-08T15:31:05Z)
• MERT: Acoustic Music Understanding Model with Large-Scale Self-supervised Training [74.32603591331718]
  We propose an acoustic Music undERstanding model with large-scale self-supervised Training (MERT), which incorporates teacher models to provide pseudo labels in the masked language modelling (MLM) style acoustic pre-training.<n> Experimental results indicate that our model can generalise and perform well on 14 music understanding tasks and attain state-of-the-art (SOTA) overall scores.
  arXiv  Detail & Related papers  (2023-05-31T18:27:43Z)
• Long-Term Rhythmic Video Soundtracker [37.082768654951465]
  We present Long-Term Rhythmic Video Soundtracker (LORIS), a novel framework to synthesize long-term conditional waveforms. We extend our model's applicability from dances to multiple sports scenarios such as floor exercise and figure skating. Our model generates long-term soundtracks with state-of-the-art musical quality and rhythmic correspondence.
  arXiv  Detail & Related papers  (2023-05-02T10:58:29Z)
• Deep Performer: Score-to-Audio Music Performance Synthesis [30.95307878579825]
  Deep Performer is a novel system for score-to-audio music performance synthesis. Unlike speech, music often contains polyphony and long notes. We show that our proposed model can synthesize music with clear polyphony and harmonic structures.
  arXiv  Detail & Related papers  (2022-02-12T10:36:52Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.